The NHLBI Division of Cardiovascular Diseases strategic plan (September 2008) highlighted the need to identify strategies to halt the AF epidemic and its associated morbidity. Recognizing novel triggers of atrial fibrillation (AF) including sleep-disordered breathing (SDB) and autonomic dysfunction is critical, as the incidence of AF is not fully explained by known risk factors. Although >1.1 million polysomnograms (PSGs) are performed annually in the US, all of which include continuous respiratory, EEG and ECG monitoring, information from these signals that may potentially identify patients at risk for future AF is not utilized. Our goal, therefore, is to identify markers from both standard nocturnal PSG scoring and from the new analyses of the PSG ECG channel that predict incident AF. Capitalizing on existing rigorously collected data (PSG including ECG monitoring, biomarkers, etc) and the multicenter infrastructure of a longitudinal cohort study (Outcomes of Sleep Disorders in Older Men Study; n~3000 participants) provides a strategic opportunity to assess the ability of PSG markers to predict individuals at increased risk for clinically adjudicated AF over an 8 year follow-up period. Although our prior cross-sectional work has shown a 2-4 fold higher odds of AF related to SDB (largely central apnea) and a relationship of SDB with both oxidative stress and systemic inflammation, prior reports have not clarified the role of PSG markers including ECG-derived markers as predictors of incident AF and have not characterized underlying linking pathophysiology. The proposed research is designed to: 1) Identify PSG-based SDB phenotypes that predict incident AF, including the relative contribution of central apnea and periodic breathing versus obstructive apnea, and explore mediation by systemic inflammation and oxidative stress, 2) Identify PSG-derived ECG markers of atrial ectopy and autonomic imbalance and evaluate their utility as predictors of incident AF and 3) Develop and test an AF clinical prediction rule incorporating significant PSG and ECG marker AF predictors. Our group is well-positioned to undertake the study aims given our expertise in epidemiology, sleep medicine, biostatistics, electrophysiology, dynamical systems and signal processing.